Private, local-only extension with Ollama
AI features that never leave the machine.
For: Privacy-first and offline-leaning developers
Point providerUrl at a local Ollama server and the extension gets AI features with no cloud call at all. The same Anouk code works against a hosted provider when the user prefers.
The problem
Privacy-conscious users won't paste sensitive content into a cloud model, and building a separate local-only path is extra work. You want one code path that works both ways.
How Anouk handles it
- ▸providerUrl is set to http://localhost:11434/v1/chat/completions in the settings panel.
- ▸Because Anouk speaks the OpenAI chat-completions shape, Ollama is a drop-in — no code change.
- ▸The user can flip between a local and a hosted model at runtime; your extension stays the same.
settings (entered by the user)
providerUrl: http://localhost:11434/v1/chat/completions
model: llama3
apiKey: (blank for local Ollama)Anouk features used
OpenAI-compatible adapterRuntime provider switchingLocal Ollama support
Build this in an afternoon
Scaffold with anouk init, drop in the call above, and point it at a provider.